flux-controller-patch-releases

Pass

Audited by Gen Agent Trust Hub on Jul 26, 2026

Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill requires the agent to run local shell commands using git and gh (GitHub CLI) to perform release tasks such as branch management, committing changes, tagging, and merging pull requests. Some commands, like git push --force-with-lease, involve destructive operations on remote repositories.\n- [DATA_EXFILTRATION]: The agent communicates with GitHub to fetch metadata and push release artifacts. While these actions are directed at the official 'fluxcd' organization repositories and are necessary for the skill's function, they constitute external network activity.\n- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it instructs the agent to read and summarize content from external pull request titles and descriptions.\n
  • Ingestion points: Metadata for pull requests is fetched from GitHub using gh pr view as specified in SKILL.md.\n
  • Boundary markers: There are no explicit instructions or delimiters provided to help the agent distinguish between its instructions and the content retrieved from PRs.\n
  • Capability inventory: The agent can perform significant actions including writing to CHANGELOG.md, creating git commits and tags, and merging pull requests on GitHub.\n
  • Sanitization: The skill lacks instructions for sanitizing or validating the PR content before it is incorporated into changelogs or introductory summaries.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 26, 2026, 11:25 PM
Security Audit — agent-trust-hub — flux-controller-patch-releases